We present M3ED, the first multi-sensor event camera dataset focused on high-speed dynamic motions in robotics applications. M3ED provides high-quality synchronized and labeled data from multiple platforms, including ground vehicles, legged robots, aerial operating challenging conditions such as driving along off-road trails, navigating through dense forests, performing aggressive flight maneuvers. Our also covers demanding operational scenarios for cameras, scenes with high egomotion...

In this work, we present a hardware-efficient architecture for pedestrian detection with neuromorphic Dynamic Vision Sensors (DVSs), asynchronous camera sensors that report discrete changes in light intensity. These imaging have many advantages compared to traditional frame-based cameras, such as increased dynamic range, lower bandwidth requirements, and higher sampling frequency power consumption. Our is composed of two main components: an event filtering stage denoise the input image...

Event-based sensors have recently drawn increasing interest in robotic perception due to their lower latency, higher dynamic range, and bandwidth requirements compared standard CMOS-based imagers. These properties make them ideal tools for real-time tasks highly environments. In this work, we demonstrate an application where event cameras excel: accurately estimating the impact location of fast-moving objects. We introduce a lightweight representation called Binary Event History Image (BEHI)...

Spectral coding is of special interest in high-sensing-capacity imaging applications. This letter presents the design, fabrication, and experimental validation a lightweight robust 3-D-printed compressive horn antenna, capable performing spectral for such First, pyramidal antenna designed to operate within E-band. Then, dielectric piece with pseudorandom geometric pattern inserted inside distort radiation both near- far-field regions. The performance antennas evaluated by their return loss patterns.

This letter presents the first experimental results of our three-dimensional (3D) millimeter-wave (mm-wave) Compressive-Reflector-Antenna (CRA) imaging system. In this prototype, CRA is 3D-printed and coated with a metallic spray to easily introduce pseudo-random scatterers on surface traditional reflector antenna (TRA). The performs pseudo random coding incident wavefront, thus adding spatial diversity in region enabling effective use compressive sensing (CS) techniques. fed...

This paper presents a novel end-to-end system for pedestrian detection using Dynamic Vision Sensors (DVSs). We target applications where multiple sensors transmit data to local processing unit, which executes algorithm. Our is composed of (i) near-chip event filter that compresses and denoises the stream from DVS, (ii) Binary Neural Network (BNN) module runs on low-computation edge computing device (in our case STM32F4 microcontroller). present architecture provide an implementation in...

This paper presents a Dynamic Vision Sensor (DVS) based system for reasoning about high-speed motion. As representative scenario we consider robot at rest, reacting to small, fast approaching object speeds higher than 15 m/s. Since conventional image sensors typical frame rates observe such an only few frames, estimating the underlying motion considerable challenge standard computer vision systems and algorithms. We present method motivated by how animals as insects solve this problem with...

This paper presents a new millimeter-wave sensing system used for near-field imaging applications. The presented is composed of parabolic reflector which fed by an array compressive antennas. A Compressive Horn Antenna (CHA) manufactured inserting dielectric piece with pseudo-random geometric pattern inside Pyramidal (PHA). CHA capable performing spectral coding, special interest high-sensing-capacity antenna fabricated using additive manufacturing (3D printing) techniques and then spray...

This paper presents our preliminary results for a three-dimensional (3D) imaging of an on-the-move target using MIMO millimeter-wave (mm-wave) radar, which uses 9 transmitters and 12 receivers. The operating frequency the mm-wave radar is from 70 GHz to 77 GHz. Experimental show that images can be created as under detection moves in front system, combined with 3D video continuous movement target. work paves way towards system used at checkpoints; thus enabling potential threats, enhancing...

We show that many perception tasks, from visual recognition, semantic segmentation, optical flow, depth estimation to vocalization discrimination, are highly redundant functions of their input data. Images or spectrograms, projected into different subspaces, formed by orthogonal bases in pixel, Fourier wavelet domains, can be used solve these tasks remarkably well regardless whether it is the top subspace where data varies most, some intermediate with moderate variability--or bottom least....

This paper presents the design and fabrication of a Compressive Reflector Antenna (CRA)for high-sensing-capacity millimeter-wave imaging applications. The CRA is fabricated using additive manufacturing or 3D printing metalized by applying silver conductive coating spray on its surface. near-fields are measured when it fed conical horn antenna, compressive perturbed cavity. singular value distribution sensing capacity aforementioned antennas calculated compared.

This paper describes a high frequency method that can be used to compute the scattered field produced by electrically large composite (multi-layered) scatterers having boundaries with arbitrary shapes. In this work, scatterer is comprised of lossy, dispersive dielectric region backed metallic region. The computes equivalent currents on scatterer, making use transmission and reflection coefficients, in order generate structure. accuracy has been validated experimental measurements vortex lens...

A major challenge in reinforcement learning is exploration, when local dithering methods such as epsilon-greedy sampling are insufficient to solve a given task. Many recent have proposed intrinsically motivate an agent seek novel states, driving the discover improved reward. However, while state-novelty exploration suitable for tasks where observations correlate well with reward, they may not explore more efficiently than approaches environments two well-correlated. In this paper, we...

A major challenge in reinforcement learning is exploration, when local dithering methods such as epsilon-greedy sampling are insufficient to solve a given task. Many recent have proposed intrinsically motivate an agent seek novel states, driving the discover improved reward. However, while state-novelty exploration suitable for tasks where observations correlate well with reward, they may not explore more efficiently than approaches environments two well-correlated. In this paper, we...

This paper presents a Dynamic Vision Sensor (DVS) based system for reasoning about high speed motion. As representative scenario, we consider the case of robot at rest reacting to small, fast approaching object speeds higher than 15m/s. Since conventional image sensors typical frame rates observe such an only few frames, estimating underlying motion considerable challenge standard computer vision systems and algorithms. In this present method motivated by how animals as insects solve problem...

This paper presents a novel end-to-end system for pedestrian detection using Dynamic Vision Sensors (DVSs). We target applications where multiple sensors transmit data to local processing unit, which executes algorithm. Our is composed of (i) near-chip event filter that compresses and denoises the stream from DVS, (ii) Binary Neural Network (BNN) module runs on low-computation edge computing device (in our case STM32F4 microcontroller). present architecture provide an implementation in...